I too have just been in my dome having a play with all of this. Here is my order of events and what happened:
Slave dome – Moves to correct sync point with mount in home position
Slew to M29 (in East) from Cartes du Ciel – Mount moves, dome moves to correct sync position
Slew to M3 (in West) from Cartes du Ciel – mount moves, dome moves to correct sync position
Slew to M29 manually from SGP – mount moves, dome moves to correct position
Centre on M27 manually from SGP – mount moves, dome moves to correct position
Park – everything parks correctly
Run sequence to ‘image’ M29 – dome starts moving, mount moves to target, dome stops in completely wrong position, then about 15 seconds later moves to correct position
Tried M29 sequence again – same as above
Slew to M29 – mount moves to correct position
Slave Dome – Dome moves to correct position
SGP Log: https://www.dropbox.com/s/mq2povmq1ceduem/sg_logfile_20191023164306.txt?dl=0
Dome Log: https://www.dropbox.com/s/kfuoepeibjbf9i7/ASCOM.Pulsar_Observatories_Dome.1643.380870.txt?dl=0
My conclusion is that this is all going wrong when in a sequence…
Reading through the logs, it looks like it is doing this on Run Sequence: slaving the dome and sending it to sync with the mount in the home position. As it is on route to the mount home position, the instruction to slew to the target azimuth arrives from the Sequence Thread. This is an incorrect value. Then the Telescope Thread kicks in with its attempt at the dome sync azimuth for the active target and comes up with ‘no adjustment needed’, so the dome carries on its journey to the incorrect position. Then Dome Thread has a go and comes up with the correct azimuth value. The Dome starts its journey to the newly established correct value. The plate solve routine kicks in, thinking that all is settled with the Dome, when actually it is moving to the correct place - the plate solve image is of the inside of the Dome and it all fails.
If the Dome is synced manually while not in a sequence, the azimuth comes from the Dome Thread and all is perfect.
Jared - does that look like it could make sense and that the azimuth is calculated from different areas, one or more of which could be doing it incorrectly?